hooman650

/

bge-large-en-v1.5-onnx-o4

like 0

How do you convert the source code to ONNX format? And what your environments?
I have tried 'optimum-cli' and 'optimum.onnxruntime.ORTOptimizer ', in [torch1.12 + optimum1.16.1 + python3.8.5].
But the onnx file I got has lack of the input name 'token_type_ids', only got the other two, which will occur errors in the load & predict process.

While the same 'optimum-cli' way I applied to convert the 'bge-rerank-large' model, only difference is task-type from 'feature-extraction' to 'text-classification', and the onnx file I got works well.

Can you show me your converting scripts and working environments? thx alot.